Probabilistic Sensor Data Fusion
in a nutshell
Prof Dr Frank Deinzer's research focuses on probabilistic sensor data fusion and indoor localization. By combining information from multiple sensors, more reliable and accurate positioning can be achieved in environments where GPS is unavailable or unreliable.
Beyond these core topics, his work extends into pattern recognition, data mining, machine learning, computer graphics, and audio processing. Publications highlight applications such as smartphone-based indoor navigation, probabilistic fusion models for activity recognition, medical image registration and volume rendering, as well as audio-based classification methods. Together, these contributions demonstrate how statistical methods and AI can transform raw sensor and signal data into meaningful knowledge for healthcare, navigation, and multimedia applications.
Current project(s)
Indoor Localization
| project title | Indoor Localization |
| summary | Indoor localization refers to the process of determining the position of people or objects inside buildings where GPS signals are weak or unavailable. It relies on technologies such as Wi-Fi, Bluetooth, RFID, or sensors to estimate location with high accuracy. This field is crucial for applications like navigation in large facilities, asset tracking, emergency response, and smart building management. Research focuses on improving precision, reducing costs, and ensuring privacy while adapting to complex indoor environments. |
| key words | probabilsitic sensor fusion; indoor localization; particle filter |
| collaborators | Future Shape GmbH, Munich & FhG IIS, Nürnberg |
| funding | HTA |
| duration | unlimited ongoing project |
| websites | simpleloc.de, www.youtube.com/channel/UClX5MbC01OcG-Cfp0vp2Oiw |
Optimization of additive printing processes
| project title | Optimierung additiver Druckprozesse durch ergänzende Anwendung von statistischer Versuchsplanung bei ML-Algorithmen zur Modellerstellung |
| summary | In additive manufacturing, experimental tests are essential for process qualification but are often costly and time-consuming. To reduce this effort, prediction and regression models are developed based on experimental data, traditionally using statistical design of experiments with polynomial regression. With increasing data availability, machine learning methods have emerged, offering flexible modeling of complex relationships. While DoE provides mathematically interpretable models and statistical validation, ML excels at capturing nonlinear interactions, with each approach compensating for the other’s limitations. The research project focuses on combining DoE and ML to improve prediction accuracy and optimize evaluation strategies in additive manufacturing. |
| key words | additive manufacturing; design of Experiments; machine learning; prediction models |
| collaborators | Hochschule Coburg/Technologie Transferzentrum Oberfranken |
| funding | Hochschule Coburg/Technologie Transferzentrum Oberfranken |
| duration | 2024-2027 |
HISÜL
| project title | HISÜL - Hochsichere intelligente Steuerung und Überwachung von Lieferdrohnen |
| summary | The project aims to create a secure and intelligent control and monitoring system for autonomous delivery drones to enhance efficiency, reliability, and reduce costs and risks. Using artificial intelligence, data science, and machine learning, the system will continuously monitor and optimize drone operations while considering legal requirements. It will be designed as a shared on-board and off-board solution, combining drone-based components with a ground station and big data infrastructure. The final product is intended for commercialization in logistics, both integrated into existing solutions and as a standalone system. |
| key words | autonomous delivery drones; control and monitoring system; artificial intelligence; logistics |
| collaborators | Emqopter GmbH; Gaitd GmbH; Technische Hochschule Würzburg Schweinfurt |
| funding | Bayerisches Verbundforschungsprogramm (BayVFP) des Freistaates Bayern |
| duration | 2024-2027 |
research group
Prof. Dr. Frank Deinzer
(Professor for Statistical Sensor Data Fusion and Indoor Localization)
Max Werner
(PhD student)
Markus Bullmann
(PhD student)
Sebastian Baader
(PhD student)
Pavel Chizhov
(PhD student)
Nisha Lakshmana Raichur
(PhD student, FhG IIS)
Valentin Wiesner
(PhD student, HS Coburg)
contact for collaboration
Prof. Dr. Frank Deinzer
frank.deinzer[at]thws.de